世界银行-CLARE：一种用于弹性估计的因果机器学习方法（英）-2026.1.pdfVIP

d

e

z

i

r

o

h

t

u

A

e

rPolicyResearchWorkingPaper11292

u

s

o

l

c

s

i

D

c

i

l

b

u

P

d

e

z

i

r

o

h

t

u

A

e

r

u

sCLARE

o

l

c

s

i

D

c

i

l

bACausalmachineLearningApproach

u

P

toResilienceEstimation

d

eTalipKilic

z

i

r

o

hMarcoLetta

t

u

A

ePierluigiMontalbano

r

uFedericaPetruccelli

s

o

l

c

s

i

D

c

i

l

b

u

P

d

e

z

i

r

o

h

t

u

A

e

r

u

s

o

l

c

s

i

D

c

i

l

b

u

P

DevelopmentEconomicsAverifiedreproducibilitypackageforthispaperis

DevelopmentDataGroupavailableat,

January2026clickherefordirectaccess.

PolicyResearchWorkingPaper11292

Abstract

Thispaperproposesanewresilienceindex,CLARE(CausalthatCLAREoutperformsexistingresiliencemetricsand

machineLearningApproachtoResilienceEstimation),alternativeapproachestopredictfoodinsecurityout-of-

whichisrootedinanimpactevaluationframeworkandsample—bothinthefuture(dynamicforecasting)andin

causalmachinelearningalgorithmsappliedtolongitudinal